--- Log opened Wed Aug 27 00:00:23 2008
00:11 -!- dbs [n=dan@pdpc/supporter/active/denials] has quit ["Ex-Chat"]
01:27 -!- Mark__T [n=Mark__T@bluemchen.kde.org] has joined #openils-evergreen
01:38 -!- miker_ [n=miker@c-76-97-57-109.hsd1.ga.comcast.net] has quit [Read error: 104 (Connection reset by peer)]
01:38 -!- sarabee_ [n=sarabee@c-76-97-57-109.hsd1.ga.comcast.net] has joined #openils-evergreen
01:38 -!- miker_ [n=miker@c-76-97-57-109.hsd1.ga.comcast.net] has joined #openils-evergreen
01:40 -!- other_miker [n=sarabee@c-76-97-57-109.hsd1.ga.comcast.net] has quit [Read error: 104 (Connection reset by peer)]
08:14 -!- kbeswick [n=kevin@8-234.lib.laurentian.ca] has joined #openils-evergreen
08:16 -!- kbeswick [n=kevin@8-234.lib.laurentian.ca] has quit [Client Quit]
08:20 -!- kgs [n=chatzill@76.48.217.216.transedge.com] has quit [Read error: 60 (Operation timed out)]
08:35 -!- gmcharlt [n=gmcharlt@adsl-68-79-9-21.dsl.emhril.ameritech.net] has quit ["Leaving"]
08:35 -!- gmcharlt [n=gmcharlt@adsl-68-79-9-21.dsl.emhril.ameritech.net] has joined #OpenILS-Evergreen
08:44 -!- Administrator [n=chatzill@75.150.13.105] has joined #openils-evergreen
08:45 -!- Administrator is now known as kgs
09:14 -!- dbs [n=dan@pdpc/supporter/active/denials] has joined #openils-evergreen
09:21 < dbs> berick: thanks for giving ephraim a nice clear answer - in retrospect, mine was not as good as it should have been
09:23 < berick> dbs: sometimes my best answers come from a lack of patience :) copy/paste/move-on
09:24 < dbs> berick: right-o; ephraim is relatively new to linux (his library is one of the eIFL pilot libraries), so thanks!
09:24 < berick> ah, cool, ok
09:28 -!- _bott_ [n=root@NAT-staff.grpl.org] has left #OpenILS-Evergreen []
09:37 -!- _bott_ [n=root@NAT-staff.grpl.org] has joined #OpenILS-Evergreen
09:38 -!- Mark__T [n=Mark__T@bluemchen.kde.org] has left #openils-evergreen []
09:45 < dbs> miker_: search does seem a bit snappier this morning - but I've conflated things by tweaking two other pgsql parms
09:48 < dbs> upped shared_buffers from 256MB to 2GB (since 8.1, the pendulum seems to be swinging towards letting postgresql keep a death grip on its own cache rather than having crazy backup/security processes mucking up the os cache)
09:49 < dbs> and upped default_statistics_target to 100 from its default of 10, to get a better sampling of the data for the planner
09:49 < dbs> so all that + GIN indexes
09:50 < dbs> can't wait to see what happens when I actually load holdings into this beat
09:50 < dbs> beast
09:55 < dbs> of course, I really need a testing harness to get objective measurements
09:55 < berick> it all comes back to tests ;)
09:56 < dbs> what a shock with me, eh? :)
09:57 < dbs> ooh, cool diacritics! http://dwarf.cs.uoguelph.ca/opac/en-US/skin/default/xml/rresult.xml?rt=subject&t=Kommunisticheskai%CD%A1a%20partii%CD%A1a%20Sovetskogo%20Soi%CD%A1uza&tp=subject&d=0&hc=1119
09:57 < berick> i like how FF shows the diacritics in the address bar
10:02 < dbs> moi aussi
10:05 -!- skmurphy_lt [n=e23@168.13.100.26] has joined #openils-evergreen
10:17 < sboyette> speaking of fonts: urxvt +sb -vb -fn 'xft:Bitstream Vera Sans Mono:pixelsize=13' -bg black -fg lightgray -cr green &
10:17 < sboyette> hawt
10:18 < kbeswick_> asl?
10:19 < sboyette> ???/undef/debian repository
10:20 < dbs> sboyette: i used to care about rxvt vs. xterm vs. gnome terminal vs. kterminal, but these days... the passion is gone
10:21 < eeevil> ETerm 4evar!
10:22 < sboyette> dbs: i'm just sayin' -- high-quality font in a unicode-aware term :)
10:22 < sboyette> uxterm probably has a similar mechanism, but i found this command line first
10:23 < sboyette> (and i never use xterm's TekTronix vector emulation anyway)
10:26 < sboyette> hëllø
11:06 < dbs> kbeswick_: prepare to meet thy doom! (I'm about to begin installing evergreen-autotools)
11:06 < kbeswick_> dbs: noooooooooOOOOO!!!!!1
11:07 < dbs> berick: on the whole setup.pl / config file database credential manipulation stuff, I think it would make a ton of sense to integrate that as a build-db.sh option
11:09 < berick> dbs: IOW, let build-db.sh take care of updating config.pl and offline-config.pl with the db params?
11:10 < dbs> berick: yeah - it seems to be the best-placed spot to do it, since that's what you're actually creating the database schema with
11:11 < dbs> maybe it becomes build-db.pl for sanity's sake
11:12 < berick> dbs: that probably is the best place, it's just a little misleading if you want to update your configs, but are not interested in building a database
11:12 < berick> which is where i usually find myself
11:12 < dbs> ah, perhaps it becomes 'eg-db-multitool.pl'
11:12 < berick> yeah, basically ;)
11:13 < berick> will this tool also be responsible for updating opensrf.xml?
11:13 < dbs> berick: right, I understand that - running "./configure --dbuser=foo --dbpass=bar --dbhost=grunt" without actually following up with making a target seems pretty durned strange
11:15 < dbs> yeah, that would be the idea. .pl would certainly be a better approach for working with XML
11:15 < berick> it would.
11:15 < dbs> although someone has undoubtedly written an XML DOM implementation in sh
11:15 < berick> heh
11:15 < berick> probably
11:17 < berick> so, the idea is, autotools does no DB user/pass/host handling.  eg-db-multitool.pl is responsible for building the DB and updating config files.  ?
11:17 < berick> if that's the case, i wonder if eg-db-multitool.pl will evolve into the generic config file fleshing utility
11:18 < dbs> heh
11:19 < dbs> in which case, keep build-db.sh as is, and just have eg-configurator.pl call it
11:20 < berick> yeah.
11:20 < berick> that certainly lowers the startup cost
11:24 < dbs> so we could start with eg-configurator (or whatever) for just db config stuff to start with, then hang more functionality onto it as we go
11:24 < dbs> settings-setter.pl :)
11:26 < berick> paper-hander-outer.pl
11:26 < berick> dbs: i think that's a good approach
11:54 < dbs> it occurs to me that we should probably have a --opensrf-user config option, and 1) sed all scripts that check for that user accordingly, and 2) automatically use the appropriate install options (ownership + privs) for installed files
12:01 < dbs> berick: I added your example to the ubuntu install instructions (already had the "delete -e #" instruction in there but the visuals definitely help). thanks again
12:07 < skmurphy_lt> dbs, kgs - just so y'all know, I've got a nearly complete document on /basic/ reporting.  It doesn't cover template creation.  You guys want it?
12:07 < eeevil> YEEEEEEEESSSSSSSSSSSSSSSSSSS
12:08 < dbs> skmurphy_lt: enthusiastic yes
12:09 < skmurphy_lt> no one has edited it yet, so i'll be happy to take any advice/critism on parts that are confusing... let me finish up the glossary section and i'll send it to you guys and some PINES folks
12:09 < skmurphy_lt> Oh, and it's got some PINES-centric stuff in it
12:11 < sboyette> you have just become kgs's favorite person
12:15 -!- phase_bb [n=phase_bb@m3f5e36d0.tmodns.net] has quit [Read error: 110 (Connection timed out)]
12:15 < dbs> a glossary? ohmygod.
12:15  * dbs pinches himself
12:29 < dbs> meh, lots of perl defunct processes on this fresh trunk-opensrf + autotools-eg install
12:29 < dbs> listeners, but no drones
12:32 < dbs> settings-tester.pl seems happy, and nothing obvious in logs with loglevel at 4. bah.
12:34 < berick> dbs: i'd check the _unix logs
12:34 < skmurphy_lt> dbs: a simple glossary; it goes over some jargon i use like 'database object' and 'operator'
12:35 < dbs> berick++
12:35 < dbs> "2008/08/27-12:18:58 Couldn't open pid file "/openils/var/pid/open-ils.actor_unix.pid" [No such file or directory]."
12:36 < berick> the last refuge for random exceptions...
12:36 < kbeswick_> dbs: ahh crap must have forgotten to change something from /openils/var/pid to /openils/var/run
12:36 < dbs> that's what I get for grepping for "ERR"
12:37 < dbs> kbeswick_: also deal with $prefix/var/sock while you're at it :)
12:38 < kbeswick_> indeed
12:39 < dbs> huzzah! Now I can go get food
12:46 < dbs> holy hell - conify works out of the box! (well, after dropping dojo-release tarball into place)
12:50  * dbs weeps at the harmonious beauty of the 1.4 universe that is falling into place
12:51 < berick> it's been a long time in the making
12:51 < dbs> btw, gfawcett is all set to roll
12:52 < dbs> the only other oddity I've noted in the autotools beast (and this might be an oddity from trunk) is that I only have one aout and one aou
12:53 < dbs> just "Consortium" and "Example Consortium". otherwise, goodness
12:54 < kgs> skmurphy_lt++ # you rock!!
12:55  * dbs peers at his oils_i18n_gettext handiwork to see if its his fault
12:57 < dbs> so kbeswick_ - think you can have /var/pid and /var/sock sorted out today, so we can merge your handiwork?
12:57 < kbeswick_> dbs: yes indeed
12:59 < dbs> also, perhaps rename "oils_config" to "eg_config" to reflect our emphasis on evergreen rather than open-ils?
12:59 < kgs> dbs++
13:04 < kbeswick_> dbs: deal
13:11 < dbs> yep, oils_i18n_gettext buggy is mine
13:22 < kbeswick_> dbs: still there?
13:22 < dbs> sadly
13:22 < atheos> what would be the quickest dirtiest way to remotely retrieve all patron e-mails from evergreen from a particular branch?  I need to "recreate" our monthly newsletter e-mail
13:23 < kbeswick_> dbs: i got an error that i have never gotten before while pushing my branch to the server
13:23 < dbs> oh
13:23 < kbeswick_> This transport does not update the working tree of: sftp://.......
13:23 < dbs> crappy
13:23 < dbs> oh, that's fine
13:23 < kbeswick_> but it still said pushed up to current revision
13:24 < kbeswick_> dbs: alright... things should be good now
13:24 < dbs> yeah, to build a tarball i had to log in, do "bzr checkout", then "bzr export foobar.tar.gz" - the checkout created a tree with all the files in it
13:40 < dbs> okay - berick, miker_, if you want to try out kbeswick_'s work you can grab it from http://142.51.8.34/bzr/evergreen/autotools.tar.gz
13:40 < dbs> I'm going to go get lunch, then come back and start prepping a merge
13:41 < dbs> so test now, or test later (when the stuff is merged) - heh
13:41 < kbeswick_> dbs: what distro did you test on?
13:41 < dbs> kbeswick_: ubuntu 7.10
13:41 < kbeswick_> dbs: im still worried about my perl tests, but if they don't work, ill just take them out or fix them
13:41 < dbs> your perl tests?
13:42 < kbeswick_> i test for the existence of certain libraries by just checking if the files are in the locations they were in on my computer
13:43 < dbs> oh, yeah, take those out
13:43 < kbeswick_> k
13:43 < dbs> it was a good try, but the location of perl modules can be highly variable
13:44 < dbs> (just looking at the tests now)
13:44 < kbeswick_> i figured
13:44 < dbs> we can perhaps get sboyette to work his magic later
13:45 < dbs> ugh, I really don't like the "find /usr -name libdbdpgsql.so" kind of check either
13:45 < kbeswick_> dbs: i think i only did that because it wouldnt find it when i checked for it the same way as all the other libraries
13:45 < kbeswick_> dbs: maybe it was becuase it wasn't in my ld path?
13:46 < dbs> kbeswick_: probably
13:46 < kbeswick_> dbs: ill try it normally again
13:46 < dbs> those libs should be in your ldpath if you followed the install directions, though (create /etc/ld.so.conf.d/eg.conf with the appropriate entries, then ldconfig)
13:47 < dbs> ldconfig -p will show you what's in the ld cache
13:48 -!- sboyette [n=mdxi@li11-97.members.linode.com] has quit [Remote closed the connection]
13:53 -!- sboyette [n=mdxi@li11-97.members.linode.com] has joined #openils-evergreen
13:57 < kbeswick_> dbs: a normal test is not finding libdbdpgsql
13:57 < dbs> kbeswick_: does "ldconfig -p | grep dbd" find it?
13:57 < kbeswick_> dbs: yes
13:58 < dbs> then autoconf sucks
13:58 < berick> heh
13:58 < kbeswick_> dbs: that it does. that it does.
13:59 < skmurphy_lt> dbs, kgs - can y'all give me your email addresses right quick?
13:59 < kgs> kgs@esilibrary.com
13:59 < dbs> skmurphy_lt: dan@coffeecode.net
14:02 < dbs> kbeswick_: http://osdir.com/ml/sysutils.autoconf.general/2004-08/msg00050.html
14:04 < dbs> kbeswick_: question - what function are you checking for in lidbdbpsql.so?
14:04 < dbs> i.e. what does your AC_CHECK_LIB line look like?
14:05 < kbeswick_> chcking for dbd_connect
14:07 < kbeswick_> dbs: i checked the source for libdbdpgsql to find a function to check for
14:07 < dbs> and the config.log just says it can't find the library, period?
14:08 < dbs> kbeswick_: yeah, nm confirms that it exists :)
14:11 < kbeswick_> dbs: it tests by linking to it, but it returns an error because it cant find it
14:11 < kbeswick_> weird
14:13 -!- skmurphy_lt [n=e23@168.13.100.26] has quit [Read error: 60 (Operation timed out)]
14:18 < dbs> kbeswick_: http://74.125.95.104/search?q=cache:H1l3ma7ovhcJ:www.polarhome.com:793/doc/libdbi-devel-0.8.3/programmers-guide/x276.html+AC_CHECK_LIB+libdbi&hl=en&ct=clnk&cd=6&gl=ca&client=firefox-a
14:19 < dbs> crazy cached link from google for how to check for a libdbi driver properly
14:25 < dbs> heh, it's also part of the standard programmer's guide for libdbi
14:25 < dbs> it's just that their web site version of the docs is out of date
14:27 < kbeswick_> dbs: haha. well, it works!
14:28 < dbs> woohoo
14:28 < dbs> one down - now that crazy libopensrf.so checking to kill
14:29 < dbs> this ain't doing it for me: "LDFLAGS=-L/openils/lib gcc -o conftest -g -O2   conftest.c -lopensrf"
14:30 < berick> LDFLAGS=-Lc:\Program Files\Evergreen
14:30 < dbs> ah, but this does:  gcc -o conftest -g -O2 -L/openils/lib  conftest.c -lopensrf
14:30 < dbs> berick++
14:32 < dbs> kbeswick_: so you could use opensrf_config to grab the libdir, pass that as LDFLAGS into configure, and that theoretically will generate -L for the lib check test
14:32 < dbs> sorry, "osrf_config"
14:32 < dbs> and now... food. for real. finally
14:32 < kbeswick_> dbs: how would you know the path of osrf_config?
14:33 < dbs> kbeswick_: uh, evergreen depends on opensrf as a prereq, right?
14:33 < kbeswick_> yeah
14:34 < dbs> I mean - if we get the user to point to osrf_config, we can auto-configure most of the dirs
14:35 < kbeswick_> that is true
14:36 < dbs> bah, let's make that a post-merge activity
14:36 < kbeswick_> right now it checks for the library where the user specifies that it should be
14:36 < dbs> yeah, you're setting opensrf-libs to /openils/lib by default, so you've got the info you need to make AC_CHECK_LIB work
14:37 < dbs> or tell us dumbasses to set LDFLAGS explicitly before invoking configure at all
14:38 < dbs> man, autotools makes things so easy
14:39 < dbs> food. for real.
14:39 < kbeswick_> sure dbs.. sure...
14:56 < phasefx> atheos: did you ever figure it out?
14:57 < atheos> phasefx, no, moved on to something else on my list :P
14:57 < phasefx> hehe
14:58 < atheos> it's an evolving list
14:58 < phasefx> select email from actor.usr where home_ou = SOME_NUMBER and email is not nullif(email,'');  ?
14:59 < phasefx> hrmm
14:59 < atheos> I'm looking to do this from a box that doesn't have direct sql access. was wondering if there is a perl solution
14:59 < atheos> API thingy
15:00 < phasefx> I think most if not all sql type queries can be made into cstore json-query, if you have access to the cstore service (the OPAC and client don't, but a perl script on a server might).  But I don't know how that works
15:01 < phasefx> nullif(email,'') is not null;  sorry (for the logs)
15:09 < dbs> kbeswick_: I has eaten
15:11 < eeevil> atheos: if you have direct OpenSRF access, yeah, you can get the emails (sorry ... calls)
15:12 < kbeswick_> dbs: good, we cannot have a malnourished dbs
15:14 < dbs> kbeswick_: rather than "find", your libopensrf test could probably be something like 'test -e ${OPENSRF_LIBS}/libopensrf.so'
15:15 < kbeswick_> dbs: ah. that works
15:19 < kbeswick_> dbs: updated.
15:25 < eeevil> atheos: hrm ... well, actually, no you can't.  that's in trunk (1.4) ...
15:26 < phasefx> could abuse crazy_search and get the data
15:26 < atheos> I'm trying to avoid doing this on our evergreen servers, because they belong to the consortia.  I need to do this from one of our boxes, remotely.
15:27 < phasefx> could run a report
15:27 < eeevil> atheos: well.. there's ... ha, yeah, reports
15:27 < dbs> dump good old csv
15:27 < atheos> hmm
15:28 < eeevil> which means I need to turn on clark kent, of course ;)
15:28 < atheos> well, i'm kinda trying to automate this in a shell script
15:28 < atheos> with as little manual intervention as possible
15:28 < eeevil> reports can be set to run on a recurring basis at, say, midnight
15:28 < eeevil> then generate the email at 5am
15:29 < eeevil> the problem, of course, is finding the correct output
15:29 < eeevil> but, reports can send email
15:30 < phasefx> oh email will be a fun.. goes to a local account with procmail, gets the URL stripped out of it, wget the url, and into another script it goes :D
15:30 < atheos> yea, that might be too rube goldbergish. although, I guess I could have it send to an e-mail that gets piped to my newsletter report
15:31 < atheos> exactly phasefx
15:32 < atheos> bah, I'll just do it from the sql box. I have access to it, an nobody will know the wiser. haha
15:32 < eeevil> :)
15:33 < phasefx> <---- just marking this spot in the IRC log... no real reason  *8)
15:33 < atheos> oh yea, these logs goes public don't they
15:36  * dbs idly wonders while reading autotools diffs if berick still responds to highfalutin at gmeal.com
15:36  * dbs also notices that the Open-ILS/admin directory still exists, and laughs an eeevil laugh
15:38 < dbs> kbeswick_: hey, man, you didn't merge in my SQL patch
15:39 < dbs> hrm, or a few other patches from the last day or two
15:39 < dbs> Now I actually have to _read_ the patch
15:54 < dbs> kbeswick_: hey little dude, you don't really want Open-ILS/xul/staff_client/Makefile.in in there, as that's a generated file, right?
16:08 -!- djfiander [n=djfiande@bas6-london14-1168070119.dsl.bell.ca] has joined #OpenILS-Evergreen
16:09 < dbs> djfiander!
16:09 < dbs> good to see you
16:09 < djfiander> dbs!
16:09 < djfiander> I'm on vacation, technically.
16:09 < berick> dbs: gmail locked that account a while back ;)
16:09 < djfiander> but I wanted to hear what kgs's voice sounded like ;-)
16:09 < berick> hi, djfiander
16:10 < kgs> djfiander: I know I wanted to hear yours :)
16:10 < kgs> or were you sans mike?
16:10 < djfiander> kgs: my mic wasn't behaving, and the call wasn't about me :)
16:10 < dbs> okay - I'm a-gonna push kbeswick_'s patch (along with my tender edits to try and avoid reverting any patches, of course)
16:11 < djfiander> the idea of downloading individual records from oclc and then placing orders sounds horribly inefficient.
16:11 < berick> dbs: wanna do the honors of destroying Open-ILS/admin?
16:11 < djfiander> exporting a list from oclc, then loading as 'selection list' might be a good workflow to explore though
16:11 < dbs> berick: I'll do that in a subsequent patch
16:11 < dbs> berick: with great pleasure
16:11 < berick> cool
16:12 < berick> ILS repo is getting thinner ;)
16:12  * berick is amazed at how quickly virtualbox runs windows, especially with a gig of ram
16:15 < dbs> berick: I'm amazed at how quickly vmware runs windows, especially with 256 MB of RAM
16:16 < dbs> I think it's mostly the lack of background antivirus and anti-spyware in my vmware version of windows...
16:16 < atheos> bah, who needs antivirus and anti-spyware in vmware windows. that's what snapshots are for
16:16 < atheos> just roll back. daily
16:17 < phasefx> these days I'm tempted to actually let malware run just to see what they do :)
16:18  * dmcmorris_esi blocks outbound port 25 from phasefx' MAC
16:20 < phasefx> hehe
16:20 < phasefx> I'd disable the virtual ethernet card first
16:30  * berick notes that Evergreen/Makefile.am is not needed
16:34 < dbs> berick: ah yes
16:34  * dbs notes that he DID NOT CREDIT kbeswick in the commit message
16:34 < dbs> dbs--
16:56 < sboyette> no one ever pays attention to that unless something is broken anyway :)
16:56 < sboyette> so you could be doing him a favor!
16:57 < dbs> sboyette: well, a fourth-year CS student who will be wanting a full-time job in 8 months probably wouldn't mind being able to point at a large commit during an interview :)
17:02 < dbs> not sure why kbeswick_ changed "sock" to "lock"; I guess he "slipped" that past
17:15 -!- dbs [n=dan@pdpc/supporter/active/denials] has quit ["Ex-Chat"]
18:03 -!- kgs is now known as kgs_away
18:03 -!- kgs_away [n=chatzill@75.150.13.105] has quit ["ChatZilla 0.9.83 [Firefox 3.0.1/2008070208]"]
18:26 -!- djfiander [n=djfiande@bas6-london14-1168070119.dsl.bell.ca] has quit ["leaving"]
19:34 -!- pinesol [n=supybot@209.168.247.198] has quit [Remote closed the connection]
20:23 -!- Netsplit zelazny.freenode.net <-> irc.freenode.net quits: rsinger, scottmcd, dmcmorris_esi, phasefx
20:25 -!- Netsplit over, joins: phasefx
20:26 -!- Netsplit over, joins: scottmcd
20:49 -!- dbs [n=dan@76-10-150-238.dsl.teksavvy.com] has joined #openils-evergreen
21:00 -!- rsinger [n=rsinger@c-67-161-232-211.hsd1.tn.comcast.net] has joined #openils-evergreen
21:00 -!- dmcmorris_esi [n=dmcmorri@dmcmorris.net] has joined #openils-evergreen
21:32 -!- kbeswick [n=kevin@208.96.112.207] has joined #openils-evergreen
21:33 < kbeswick> dbs: i see that you have committed my life's work.
21:45 < dbs> indeed, without giving you any credit
21:45 < dbs> dbs--
21:46 < kbeswick> haha
21:46 < kbeswick> oh well
21:47 < kbeswick> my name is on the files that matter
21:47 < kbeswick> dbs: and about the 'lock' thing, didnt we say to change sock to lock because of the FHS?
21:47 < kbeswick> or is they completely different things?
21:48 < kbeswick> are they(
21:48 < kbeswick> hahah
21:55 < dbs> kbeswick: different things
21:55 < kbeswick> gah
21:55 < dbs> I think you're looking for /var/run/evergreen/foobar.sock (according to http://www.pathname.com/fhs/pub/fhs-2.3.html#VARRUNRUNTIMEVARIABLEDATA)
21:56 < dbs> "System programs that maintain transient UNIX-domain sockets must place them in this directory."
21:56 < kbeswick> ah
21:56 < kbeswick> it may be like that for osrf too
21:57 < dbs> indeed
21:58 < kbeswick> and the xul client Makefile.in is a mistake
21:58 < kbeswick> it might be the old one, or something
21:58 < dbs> yeah, I didn't commit that
21:58 < kbeswick> godo
21:58 < dbs> godot? I've been waiting for him.
21:58 < kbeswick> and i forgot that i hadnt merged my branch for a couple of days too
21:59 < dbs> dinnae worry, i think it's all good
21:59 < kbeswick> yay, i am in view of moving on to doing something else!
22:01 < kbeswick> packaging, i'm assuming, comes after this
22:01 < kbeswick> in my plan moulded by the great miker_
22:02 < miker_> is that great as in "the great and terrible Oz"?  ;)
22:02 < dbs> or you could help write settings-setter.pl
22:03 < miker_> "pay not attention to the man behind the curtain!"
22:03 < dbs> "He's had way too much GIN!"
22:03 < miker_> indeed
22:03 < miker_> dbs: well, so have you after 1667 minutes of it :(
22:04 < dbs> miker_: I still suspect something's borked on our server
22:04 < dbs> must be raid 5
22:04 < miker_> how big is the index?
22:04 < miker_> pg_relation_size('indexnamehere')
22:04 < dbs> one second
22:05 < dbs> gotta get my vpn going
22:05 < miker_> (I'm sure you knew that, of course)
22:06 < dbs> you flatter me - I would have gone relfilenode via pg_class and du -hs relfilenode*
22:06 < dbs> <-- still a postgresql rookie
22:10 < dbs> apparently the indexes don't exist
22:11 < miker_> ....
22:11 < lisppaste6> dbs pasted "cmon you index sizes!" at http://paste.lisp.org/display/65986
22:11 < miker_> schema qualified?
22:12 < miker_> by default the index lives in the same schema as the table
22:13 < dbs> authority index is 435 MB
22:13 < miker_> that's a ton smaller than the gist version
22:13 < miker_> how many rows in authority.full_rec?
22:14 < miker_> for comparison with one o' my data sets
22:14 < dbs> ahhh
22:14 < dbs> 'metabib.metabib_keyword_field_entry_index_vector_gin_idx' = 1664 MB
22:15 < dbs> counting...
22:16 < dbs> schema qualifying helped, of course. I would swear that I did that first time around, but my eyeballs are bleeding
22:16 < dbs> 12294980 rows in authority.full_rec
22:17 < miker_> 12M ... min has 7M and is 337MB, so that's about right
22:18 < miker_> hrm... your keyword index seems big to me
22:18 < miker_> ~1.9M records for me is 399MB
22:19 < dbs> we do have some very rich records
22:20 < miker_> but an inverted index should be tied to the number of unique terms, not total terms (as gist is) :(
22:20 < lisppaste6> dbs annotated #65986 with "schema qualification required, of course" at http://paste.lisp.org/display/65986#1
22:21 < dbs> and two languages
22:21 < miker_> heh  ... I was just using (pg_relation_size(...)/1024)/102
22:21 < miker_> 4
22:21 < dbs> so - double the number of records, and double the number of terms
22:22 < dbs> (even though french is close to english, so that doubling is false cause lots will get stemmed down)
22:24 < miker_> well, I can believe it, I guess
22:24 < miker_> my 400MB index is gist, though
22:25 < miker_> anyway, any difference in "feel"?
22:29 < lisppaste6> dbs annotated #65986 with "All the index sizes and their row counts" at http://paste.lisp.org/display/65986#2
22:29 < dbs> it does feel pretty snappy, although like I said this morning, I also bumped shared_buffers to 2GB and default_statistics_target to100
22:31 < dbs> I regret not capturing the stats on the gist indexes
22:32 < dbs> oh well, I'll be building them a few more times before the year is through
22:32 < dbs> miker_: did I tell you how thrilled I was to see conify "just work" with a plain install of osrf trunk and autotools eg (and an extracted dojo tarball, of course)
22:34 < miker_> oh good!
22:35 < kbeswick> dbs: what had to be done before to make it work?
22:36 < dbs> oh, mostly monkeying around with apache configuration, praying that the python opensrf-over-http implementation worked properly, squeezing local diffs out of miker_...
22:36 < kbeswick> lolz
22:38 < miker_> :)
22:53 -!- sarabee__ [n=sarabee@c-76-97-57-109.hsd1.ga.comcast.net] has joined #openils-evergreen
22:53 -!- miker__ [n=miker@c-76-97-57-109.hsd1.ga.comcast.net] has joined #openils-evergreen
22:54 -!- sarabee_ [n=sarabee@c-76-97-57-109.hsd1.ga.comcast.net] has quit [Read error: 104 (Connection reset by peer)]
22:54 -!- miker_ [n=miker@c-76-97-57-109.hsd1.ga.comcast.net] has quit [Read error: 104 (Connection reset by peer)]
22:55 < dbs> ah, and 96183052 rows in metabib.full_rec
23:11 -!- kbeswick [n=kevin@208.96.112.207] has quit ["Ex-Chat"]
23:11 < miker__> dang...
23:11 -!- miker__ is now known as miker_
23:12 < miker_> btw, my toobs locked up after kbeswick said 'lolz' ... so, wasn't ignoring if you were talkin' at me
23:13 < dbs> nothing more than the mfr row count
23:14 < dbs> i was (gasp) doing some php work
23:29 < berick> dbs: did you guys ever run into this when using bzr / svn?  https://bugs.launchpad.net/bzr/+bug/205579
23:31 < dbs> berick: no... what are you doing to provoke it?
23:33 < berick> i have a bzr repo which was created with : bzr init-repo --rich-root-pack my-repo
23:33 < berick> followed by a bzr checkout svn://blah (local test svn repo)
23:34 < berick> so, then i try to run bzr server on the bzr repo.  trying to checkout that bzr served repo creates the error
23:34 < berick> not sure if I explained that very well :)
23:34 < berick> svn -> bzr mirror -> local checkout of bzr mirror explodes
23:37 < berick> is this ^-- the basic architecture you guys are using?   a local bzr mirror of the ILS repo
23:37 < berick> you checkout of a copy of the bzr repo, do your stuff, commit to the bzr repo, then merge changes back to the ils?
23:39 < dbs> berick: ah, I never init-repo, I just "bzr co svn://source target-dir"
23:39 < dbs> And I don't run bzr server
23:40 < dbs> we use bzr over ssh, or bzr over http
23:41 < berick> ah, ok.  good to know
23:41 < dbs> bzr co http://bzr.coffeecode.net/ILS/trunk brz-ILS-demo
23:41 < berick> running bzr server was the lowest barrier to entry for testing...
23:41  * berick looks at http
23:42 < berick> thanks, dbs!
23:42 < dbs> I know bzr server is supposed to improve performance and add some more features, but it sounds like it adds some bugs too
23:48  * dbs reads that as "thanks for wasting my time, dbs!" :)
23:50 < berick> no, you simplified the overall process, which feels a lot better
23:52 < dbs> berick: yeah, well...
23:53 < lisppaste6> dbs pasted "pride goeth before the fall" at http://paste.lisp.org/display/65994
23:54 < berick> mm, good times
23:58 < dbs> hmm. could be because I'm using bzr 1.5 on coffeecode.net, and 1.3 on my laptop
23:59 < dbs> (lenny vs hardy)
--- Log closed Thu Aug 28 00:00:37 2008